Enabling basic metrics in Azure

[kalousd]

Hi,

Terraform Version
0.9.11

I am willing to enable basic metrics while creating a azurerm_virtual_machine, however, no such argument is available (unless I am missing something). I am able to enable boot_diagnostics without any problems. See screenshot:

Would be cool to have this argument available sooner or later.

Thanks!

[tombuildsstuff]

Hey @nKaloussian

Thanks for opening this issue :)

I've taken a look into this issue and from what I can see checking the box in the Portal actually installs a VM Extension on the machine.

I've attempted to replicate this in Terraform - however I'm having issues with the VM Extension itself not installing as expected; as such I've opened this issue requesting more information on this functionality - and I'll update here when I hear back.

Thanks!

[tombuildsstuff]

Hey @nKaloussian

I've spent some time looking into this - and as I previously assumed this can be configured through a VM Extension.

For Linux Virtual machines, the LinuxDiagnostic extension in Microsoft.OSTCExtensions can be used - there's an example here. For Windows Virtual Machines the IaaSDiagnostics extension in Microsoft.Azure.Diagnostics can be used - there's an example of it's use here.

Would you be able to take a look and see if that provides the functionality you're looking for?

Thanks!

[dominik-lekse]

@tombuildsstuff Thanks for your research on this. In my opinion, setting up these extensions properly is overcomplicated. In practise, you would need these infrastructure monitoring metric for all machines.

What do you think about integrating a simple switch for the basic diagnostics as the portal does, so the the virtual machine terraform resource does the rest in the background?

[tombuildsstuff]

@dominik-lekse from a UX standpoint I'd completely agree, but I think we're going to need a change in the Compute API to support that.

Whilst we could potentially add a flag to the azurerm_virtual_machine resource to install a VM Extension; this isn't ideal as Terraform would need to store up-to-date (basic) configurations for each OS type - in addition users submitting their own VM Extension for Azure Monitoring who opt to use a different name for the VM Extension would see errors from Azure.

As such I think for the moment we're better not adding this flag unfortunately - at least until the API allows a route forward here. I've opened an issue on the Rest API Specs repository to request this feature is added to the Compute API here - which I'd recommend watching.

In the mean time - given I don't think there's an immediate path forward here I'm going to close this issue - however should an API become available we'll take another look :)

Thanks!

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. If you feel I made an error 🤖 🙉 , please reach out to my human friends 👉 hashibot-feedback@hashicorp.com. Thanks!